Professor Paul Fenn gives some of the theoretical explanations of the general insurance underwriting cycle.

In competitive markets, it is usual to assume that prices reflect the cost of production together with a reasonable level of profit, sufficient to keep suppliers in the business. For general insurers, the 'cost of production' arising from a year's underwriting could be seen as the anticipated cost of settling claims from the policies written in that year, together with the expenses associated with selling policies, managing claims, etc.

In principle, competition should ensure that premium income from these policies was just sufficient to cover these combined costs together with a 'normal' profit margin. If this were to be the case, however, it would be difficult to explain why cycles in underwriting profits emerge as they do. Given rational expectations, insurers' premium income would vary in relation to costs only insofar as there were random and unpredictable shocks to their anticipated claims experience. Consequently, research on this issue has attempted to produce a consistent theoretical explanation for the emergence of insurance cycles, and to test these theories against data on underwriting profits. The principal alternatives can be summarised under four headings.

Structure fluctuations

Early commentators on the underwriting cycle suggested that it was caused by periods of "destructive competition" followed by oligopolistic coordination to cut back supply in order to increase prices before the next round of competition arrives (Stewart, 1981). In effect, it was argued that dynamic changes in the degree of market competition caused the profit margin to fluctuate cyclically. However, as a theory this argument lacks coherence: it offers no explanation as to why insurers choose to compete and collaborate at particular points in time generally six years apart. It seems more likely that these commentators were pointing to the symptoms of the cycle rather than the underlying causes: as underwriting profits increase, new entrants will be drawn in, and the converse when losses are made. Any attempts made by one of these new entrants to increase market share through setting premiums below expected claims cannot be sustained for long in such markets without solvency problems emerging.

Forecasting errors

Venezian (1985) suggested a different explanation for the failure of premiums to behave as predicted by the theory of competitive equilibrium.

He argued that the way in which insurers estimate expected claims was not in accordance with the assumption of rational expectations. 'Naive' actuarial forecasting procedures such as the chain ladder method can work to add a degree of persistence to expectations about future claims experience.

Critics of Venezian's arguments stress the fact that insurance cycles are observed in virtually all countries, irrespective of the techniques used to estimate reserves. Moreover, his theory assumes an unrealistic degree of irrationality on the part of insurers who fail to learn from experience of past forecasting errors.

Costs and reporting lags

In an influential paper, Cummins and Outreville (1987) objected to explanations based on irrationality and/or lack of competition in insurance markets.

They argued that only theories based on competitive equilibrium and rational expectations could be persuasive. They suggested that the pricing behaviour of insurers is consistent with both assumptions, and that in the absence of adjustment costs and reporting lags there would be no discernable cycle in underwriting profits.

To begin with, however, the premiums set for current policies are based on the most recent claims experience observed by the insurer, which is inevitably available only with a lag, given the delays in recording and analysing the data. Moreover, because policies are generally sold with annual terms and premiums are revised annually, there can be a substantial lag between the setting of a new rate and its application to a particular policy. The combined effect of delayed access to claims experience and the delayed application of current rates to new policies means that the premiums associated with policies written in a given year may well reflect claims experience from the previous year. Moreover, data are not always easily available for premiums written in a given year. What is typically observed in the calendar year accounting aggregates reported to the regulator are the premiums earned within a given year; that is, they include premiums arising from policies issued up to one year prior to the accounting year.

Cummins and Outreville argued that this effect could add a second order lag to that resulting from the rate-setting lag.

The possibility of information and reporting lags between premiums and claims experience does not in itself produce a cycle in underwriting profits(1).

Cummins and Outreville further assume that claims experience is serially correlated over time due to the fact that loss shocks have a permanent component (some part of the shock is believed to have an effect on all subsequent losses). This assumption, together with the lagged relationship between premiums and claims experience, produces an autoregressive process in reported underwriting profits, and the result is an apparent cycle which is nevertheless consistent with rational expectations in a competitive market.

Capital constraint models

The Cummins and Outreville theory relies on information lags in conjunction with claims persistence to generate an observed cycle. Implicitly it assumes that both insurance and capital markets are otherwise operating perfectly.

Specifically, it assumes that financial capital adjusts instantaneously and at no cost so that premiums are perfectly elastic with respect to a given expected loss. Winter (1988, 1991)(2) argued that this assumption was unrealistic, and resulted in a failure to explain all the salient features of the insurance cycle. He maintained that, even in the absence of regulation, insurers would hold equity capital in order to make insurance contracts credible given limited liability(3). Regulatory requirements reinforce this constraint. If equity capital could nevertheless be adjusted costlessly, common supply shocks(4) could be accommodated such that prices remained at their equilibrium levels. However, adjusting external equity is costly for a number of reasons (summarised by Winter), and therefore firms choose to hold a desired level of excess internal capital in order to reduce the risk of having to raise capital externally. This induces a degree of persistence to underwriting profits in hard markets as insurers respond to common unanticipated capital shocks by keeping prices high subsequently in order to restore their net worth. Similarly, low profits persist in soft markets as the restoration of surpluses permit premiums to stay low.

Underwriting profits are therefore characterised by an autoregressive process, but in this case this is caused by a response to slowly adjusting capital rather than persistence in claims.

Side explanations

Some commentators have suggested that the capacity constraint model is incomplete in that it does not provide an explanation for how the desired level of excess capacity is determined. Subsequent models incorporate default risk endogenously by viewing insurance pricing as analogous to the pricing of risky corporate debt (Doherty and Garven, 1986). Cummins and Danzon (1997) put forward a model in which the relationship between the price of insurance and capital is derived from policyholder demand for financial quality (avoidance of insolvency risk). Consequently, a prediction of this model is that there is a long-run relationship between surplus and price: firms with lower insolvency risks can set higher premiums for the same cover. More recently Lai et al (2000) have suggested that loss expectations on the demand side of the market can also work to amplify the cycle.

Each of the theories reviewed above has, to a greater or lesser extent, been subject to empirical testing by researchers attempting to discriminate between them(5). The consensus would seem to be that there is indeed an apparent underlying cycle in general insurance profitability with a cycle length in the region of six to eight years. This cycle remains evident even after allowing for the influence of fluctuations in industry capital and competition. The implication seems to be that an insurance underwriting cycle would exist to some extent even in a competitive market with perfectly flexible access to capital. Its origin presumably lies in the adjustment lags between expected claims and premium income in conjunction with a fundamental degree of persistence in the former over time. As this persistence is arguably the consequence of the need to forecast claims experience, it is likely to be a feature of most general insurance markets, and this explains why the phenomenon is so widely observed.

NOTES

(1) If claims are random and independent across periods, then premiums set equal to lagged claims will be similarly random; the difference between two random variables will also be random.

(2) See also Gron (1994).

(3) Rational consumers would be unwilling to buy insurance from providers who are at significant risk of insolvency.

(4) Due either to industry-wide unanticipated losses or to shortfalls in investment returns, for example.

(5) See for example Choi, Hardigree and Thistle (2002).

REFERENCES

Choi, S, Hardigree, D, and P Thistle, 2002, The Property/Liability Insurance Cycle: A Comparison of Alternative Models, Southern Economic Journal, Vol 68 (3), 530-48.

Cummins, D and P Danzon, 1997, Price, Financial Quality and Capital Flows in Insurance Markets, Journal of Financial Intermediation, Vol 6, 3-38.

Cummins, D and F Outreville, 1987, An International Analysis of Underwriting Cycles in Property-Liability Insurance, Journal of Risk and Insurance, Vol 54, 246-262.

Doherty, N and J Garven, 1986, Price Regulation in Property-Liability Insurance: A Contingent Claims Approach, Journal of Finance, Vol 41, 1031-1050.

Gron, A, 1994, Capacity Constraints in Property-Casualty Insurance Markets, Rand Journal of Economics, Vol 25, 110-127.

Stewart, B, 1981, Profit Cycles in Property-Liability Insurance, in Issues in Insurance (J Long (ed)) American Inst for Property-Liability Underwriters, Malvern.

Venezian, E, 1985, Ratemaking Methods and Profit Cycles in Property and Liability Insurance, Journal of Risk and Insurance, Vol 52, 477-500.

Winter, R, 1988, The Liability Crisis and the Dynamics of Competitive Insurance Market, Yale Journal on Regulation, Vol 5, 455-499.

Winter, R, 1991, Solvency Regulation and the Property-Liability "Insurance Cycle", Economic Inquiry, Vol 29, 458-471.

- Professor Paul Fenn is the Norwich Union Professor of Insurance at the Centre for Risk and Insurance Studies at the Nottingham University Business School.

Topics